On 29 October 2025, Kazakhstan’s Parliament adopted the country’s first Artificial Intelligence (AI) Law, which comes into effect on 18 January 2026. This landmark legislation aims to balance technological innovation with ethical standards and intellectual property rights.

Kazakhstan’s copyright framework is built on both national legislation and international agreements, including the Berne Convention, the TRIPS Agreement and the WIPO Copyright Treaty. Domestically, the Civil Code and the Law on Copyright and Related Rights (1996, as amended) define the rules of authorship, ownership and protection.

Kazakhstan's newly adopted AI Law does not directly amend the Copyright Law but complements it by introducing provisions for key AI-related areas where the copyright law was silent.

Key AI Law features

• Definitions: The law provides precise definitions of terms such as artificial intelligence, AI system, AI model, machine readable format, and related concepts.
• Legal framework: The law establishes a comprehensive legal and organisational foundation for AI governance, emphasising transparency, safety and accountability in AI development and deployment.
• Developer duties: Owners and holders of AI systems must manage risks across the lifecycle, ensure safety and reliability, protect against unauthorised access, maintain documentation, and support users. Liability for harm caused by AI systems is governed under the Civil Code of Kazakhstan.
• Trusted systems: Government bodies may establish voluntary “trust lists” for high risk AI systems to promote transparency, best practices, and public confidence. Inclusion in these lists requires an audit conducted by the system’s owner or holder, but participation is optional rather than mandatory.
• Risk based regulation: AI systems are classified into different risk categories, including minimal, medium, or high risk based on their potential impact on safety and society. High risk systems face stricter oversight, while harmful applications are prohibited.
• Prohibited uses: The law bans manipulative or subconscious techniques, exploitation of vulnerable groups, social scoring (except where legally permitted), unlawful biometric data processing, and emotion recognition without consent, ensuring protection of fundamental rights and dignity.

Key copyright provisions (Article 23 of AI Law)

• Authorship: Copyright protection is reserved for works of human authorship. Works created entirely by AI systems, without human contribution, are excluded from protection.
• Prompts: Textual prompts that demonstrate human intellectual or creative effort may be recognised as copyrightable works. This raises questions about the boundary between human authorship and autonomous machine generation.
• Use of works for AI training: The use of copyrighted works for training AI models is treated as a distinct category of use, separate from traditional exploitation rights. Under the law, authors or rights holders may expressly prohibit such use in a machine-readable format—a novel mechanism that effectively resembles the creation of a new form of exclusive right.
• Legal framework: This new exclusive right is not yet recognised in the Civil Code or Copyright Law, but the AI Law signals movement in that direction.
• Mandatory labeling: The AI Law introduces mandatory labeling for AI-generated services, products, or content (images, video, audio, text) in both machine-readable and visual forms to ensure transparency and prevent misuse.

Since the level of human creativity required to establish authorship has not yet been defined, and practical mechanisms for enforcing prohibitions expressed in a machine readable form are still lacking, achieving a balance between innovation and the protection of authors’ rights will demand continued collaboration among lawmakers, courts, and industry stakeholders.